Expected Goals (xG) is an essential measure in football analytics that quantifies the probability of a shot leading to a goal based on various contextual factors such as distance, angle, and assist type. Traditional xG models often focus solely on core parameters, such as distance and angle to the goal, while overlooking important auxiliary factors like psychological, player-specific attributes, and team dynamics. This research aims to develop enhanced xG models by analyzing 257,800 shot events from top European leagues, including the Premier League, La Liga, and Serie A, between 2014 and 2023, utilizing various machine learning models. The inclusion of novel features enhanced the models’ predictive accuracy, demonstrating superior performance compared to traditional xG metrics like those provided by Understat. This research demonstrates that ensemble methods integrating LightGBM, CatBoost, and XGBoost outperform other machine learning models in predicting xG, achieving a ROC AUC of 0.81, Brier Score of 0.076, and Log Loss of 0.271, surpassing industry benchmarks and existing literature. These improvements offer clearer insights into match outcomes and player efficiency, providing coaches, analysts, and players with more effective tools for performance optimisation.